Six-month retention and changes in quality of life and substance use from a low-threshold methadone maintenance therapy programme in Durban, South Africa

Prior to conducting the study, a multi-stakeholder task team was established to inform protocol development and to oversee implementation. This included representatives from the national and provincial Departments of Health and Social Development, police, academics, harm reduction practitioners and (potential) study participants. The task team met bi-annually to reflect on progress, develop opportunities for collaboration, identify referral networks and reflect on challenges and future endeavours.

People were informed about methadone prescribing and the programme through organisations providing harm reduction and other health and social services for those who use drugs in the city, as well as through members of the community of people who use drugs. Members of the study team pre-screened potential participants to ensure that they complied with the inclusion criteria stated above.

As part of screening, a study team member collected baseline data on sociodemographic characteristics (sex, age, race, living situation, criminal record) and substance use (using WHO ASSIST and questions around current and previous drug use patterns, including opioid agonists). The study nurse conducted a medical history, physical examination, urine analysis for opioids, and an electrocardiogram (ECG). ECGs were included to identify people with potential cardiac problems at baseline and to gather data on changes over time to generate local safety data. The timing of ECGs was done following the South African guidelines for the management of opioid dependence developed by the South African Addiction Medicine Society [40], which recommends baseline, one month and annual ECG assessment. People with abnormal ECGs were referred to the nearest public hospital for further assessment. A study team member also offered HIV counselling and testing. A multi-disciplinary study team (involving clinical staff, researchers and psychosocial service staff) assessed participants’ suitability and eligible participants were medically assessed by a doctor. Participants signed a treatment contract outlining the nature of the study and the expectations of the participant and staff. The doctor initiated participants onto methadone at an appropriate dose in accordance with the South African Addiction Medicines Society guidelines, which recommends a starting dose of between 10 and 30 mg [41]. The nurse then completed the SF-12 quality of life assessment. This generic preference-weighted health outcomes assessment includes 12 questions. The Quality Metric’s Health Outcomes scoring software enables the generation of quality of life scores [42]. The SF-12 questions around physical functioning, physical state, bodily pain and general health, are used to develop a physical health composite score. Questions about vitality, social functioning, emotional state and mental health are used to develop a mental health composite score. A SF-6D Health Utility Index score is also generated. This preference-based index is used to measure health status, ranging from worst health state (0.0) to perfect health (1.0). The SF-6D score can be used to calculate quality adjusted life years for economic evaluations. The SF-12 has been validated for use in the South African context [42].

Initially, all participants came to the centre for daily observed dosing and were seen by the project nurse and other staff. Participants saw the medical doctor for clinical assessments every 3 to 7 days for the first 2 months as doses were increased and then stabilised, with no ceiling. Participants who had been on a stable maintenance dose for around 3 months were considered for take home doses with monthly medical assessments. A multi-disciplinary team involving the medical doctor, nurse and psychosocial support team decided on the commencement of take home dosing. Take home doses were terminated and daily observed doses reinstated if the participant repeatedly diverted, sold or otherwise misused the methadone supplied. Participants returned to daily observed dosing if their lives had been significantly destabalised. The SF-12, ASSIST, and HIV testing were repeated after 6 months on methadone.

All participants were provided with optional psychosocial support services. Throughout the demonstration project, a social worker and counsellor were available for individual consultations and for group sessions. Following a harm reduction approach, beneficiaries were asked to define their goals and were encouraged to talk without fear of victimisation or judgement of their drug use (past and present). Abstinence was not a requirement for receiving methadone or participation in the study. Shortly after the programme began, beneficiaries started their own peer-led support group. The social worker and/or counsellor assisted in the initial meetings, but fairly quickly the peer-led groups operated independently of ‘professional intervention’. Participation in at least two group sessions and at least one individual session each month was considered regular attendance. A group for family members and other people supporting participants on methadone was established and met every 4 months.

Weekly multi-disciplinary meetings of clinical, psychosocial, support and research staff were held to discuss participants. Clinical and psychosocial plans were developed as required.

Participants who voluntarily terminated their participation were counselled and supported through a down titration process. These participants were viewed as having exited the programme. The protocol allowed for involuntary termination if participants were repeatedly found to divert or traffic methadone. Participants were considered lost to follow-up if they missed 30 consecutive doses. Participants were able and encouraged to restart methadone at any stage if they were deemed lost to follow-up and were interested in re-entry, following a medical and psychosocial assessment.

Clinical notes and ASSISTs were completed on paper and captured into a password protected excel database. The SF-12 s were captured into Quality Metric’s Optum PRO CoRE Health Outcomes software (Eden Prairie, Minnesota) and exported into a password protected excel document. Data was merged and imported into Stata v14.2 (College Station, Texas) for analysis.

Descriptions of baseline data included proportions for categorical variables and measures of central tendency and dispersion (medians and inter-quartile ranges) for numerical variables. Substance specific and overall substance use scores were obtained from the ASSIST. Mental composite, physical composite and SF-6D scores were outputs from the Optum PRO CoRE Health Outcomes software. Paired t-tests (for normally distributed data) and the Wilcoxon signed rank test (for non-normally distributed data) were used to assess changes in substance use (ASSIST) and quality of life (SF-12 composite scores). McNemar’s test was used to calculate p values for paired data derived around injecting frequency in the preceding 3 months among participants completing baseline and 6-month ASSISTs. Retention in the study was used as a proxy measure for a positive outcome [43, 44].

Recruitment started in August 2017, and of the 110 people pre-screened 61% (67/110) were potentially eligible. Of these, 81% (54/67) signed treatment contracts and were enrolled. We initiated a total of 53 (48% of those pre-screened) people onto methadone (see Fig. 1). All participants who were contacted agreed to be pre-screened.

An overview of sociodemographic and substance use characteristics at baseline is provided in Table 1. The majority of the beneficiaries were young, black African males. Approximately half reported a criminal record and had used drugs for more than 10 years, respectively.

At baseline, nine participants reported injecting heroin within the previous 3 months while the remainder smoked heroin, usually in combination with tobacco and/ or cannabis. The majority (57%) used heroin more than four times per day. The median baseline ASSIST score for all substances was 81.5 (IQR 67–93), for opioids it was 37 (IQR 33–39) and cannabis 12.5 (IQR 3–23) (see Table 2). Cannabis (32%) was the second most common substance used.

In the month preceding initiation, seven (13%) of the participants reported using methadone, with two of them reporting that they stopped due to cost. Among them the median dose was 20 mg (IQR 10–50 mg) and reported dosing frequency ranged from once (n = 4) to thrice daily (n = 1). Additionally, six (11%) people reported using buprenorphine during the month before initiation with a median dose of 6 mg (IQR 4–8 mg) with daily (n = 3) or twice daily (n = 3) dosing.

The baseline quality of life physical composite score was 55.5 (IQR 47.5–59.3), mental composite score 41.4 (IQR 35–47.3) and SF-6d-R2 0.657 (IQR 0.630–0.738) (see Table 3). The HIV prevalence at baseline was 13% (7/54). No abnormalities were seen on baseline ECGs done on eligible participants.

Of the 110 applicants, 53 were initiated on methadone (one client signed consent but did not start methadone), with the median dose of methadone being 30 mg (IQR 30 mg). After 6 months, 81% (44/54) remained in the study. Seven people dropped out and could not be traced and two voluntarily exited. Of the nine people who started methadone and were not retained, 78% (7/9) left within 3 months of initiation. The median methadone dose among participants retained after 6 months was 55 mg (IQR 40–70 mg). The median methadone dose among participants lost to follow-up was 50 mg (IQR 30–60 mg). At the end of the 6-month period, 57% (25/44) were on daily observed therapy at the centre, while the rest received take-home doses. The median number of missed doses among participants remaining on OST after 6 months was 10 (IQR 3–21).

Voluntary group and individual psychosocial services were offered and 74% (40/54) of the participants regularly attended group sessions, 78% (42/54) participated in one or more individual sessions per month with the social worker or counsellor, and 74% (40/54) participated in group and individual sessions.

There was a statistically significant increase in the mental composite score after 6 months (p = 0.0254) among participants retained who completed an SF-12. Changes in the physical composite and SF-6d-R2 scores at 6 months (Table 3) were not significant. No new HIV infections were recorded among beneficiaries retained in the programme at the 6-month time point.

Changes in ASSIST scores between baseline and at 6 months are shown in Table 2. Among those retained, there was a notable reduction in overall substance use. Overall, ASSIST scores decreased from a median of 81.5 to 67 (p = 0.0052). Opioid ASSIST scores reduced significantly from 37 to 9 (p < 0.0001), while cannabis increased significantly from 12.5 to 21 (p = 0.0003). There were insignificant increases in reported use of cocaine, and alcohol at 6 months. Among the 13 people who did not complete the 6 month ASSIST, 9 reported to have never injected at baseline.

One patient, without a history of injecting drugs, who was living with HIV was lost to follow-up and died as a result of TB infection. One client had an opioid related overdose requiring supportive treatment. Two participants had mild increases in their QTC interval measurements after one month, both of which normalised after dose reduction and counselling.

Retention is considered the best proxy indicator for positive outcomes in the resolution of opioid use disorders [43]. Internationally, retention rates for methadone maintenance therapy differ widely. At 6 months, rates of between 3 and 88% have been identified in the literature [45]. It has been suggested that the benchmark for methadone maintenance therapy is 50% [46]. However, a 2014 review of medication for opioid use disorder treatment programmes in lower and middle income countries reports average retention in methadone programmes as 72% (range 64–80%) at 6 months and 56% (range 46–68%) at 12 months [47].

The results from our study show 6-month retention rates that are at or above retention rates found in similar settings. However, the results are not totally unexpected. The programme was designed to include a number of factors that are predictive of high retention rates such as stable accommodation and a support person. Studies have repeatedly shown that individual characteristics are not predictive of outcomes, and that contextual factors including programmatic and social factors are predictive of positive outcomes. Programmatic predictors of high retention rates include individualised flexible dosing with no ceiling dose [48], accessibility (including cost and geographical position) [49] and therapeutic relationships [50]. Social factors include peer and family support [51]. The potential of take home doses also increases positive social factors [52]. These factors were included in the study design. Most of these features are common to low threshold programmes, which show higher levels of retention than higher threshold services [17‐19].

One of the most consistent predictors of retention is methadone dose [53‐55]. The globally recommended daily dose range for methadone is 60–120 mg [11]. Interestingly, the median methadone dose among participants retained at 6 months in this study was 55 mg. This could suggest that most beneficiaries did not require very high doses of methadone, most likely because the majority were not injectors [56]. The impact of lower doses was, we believe, mediated by the notable proportion of people on flexible and patient centred dosing (home dosing). Other potential factors include the option for re-initiation after being lost to follow-up and the employment of a non-punitive approach to service delivery.

Positive change in mental health was significant at the 6-month mark. We infer that this is the result of connections with significant others, including the multi-disciplinary project team, peers and family members. These kinds of connections have been found to reduce feelings of stigmatisation and marginalisation and allow for life normalisation [57, 58]. While the analysis did not indicate any notable change in overall physical health, it is useful to bear in mind that there was low HIV prevalence at baseline, which did not change after 6 months.

The notable drop in the overall ASSIST score across all substance classes was to be expected and is supported by decades of research on the effectiveness of methadone [16, 43, 44, 48, 53, 59‐64]. There was a significant increase in the use of cannabis amongst participants, however the reasons for this increase was not investigated. While some studies suggest that cannabis use is a predictor of lower retention rates [65, 66], a 2009 review concludes that it is not associated with worse opioid use outcomes. The review notes that there is no need to address cannabis use during methadone maintenance therapy unless the service user requests assistance [67]. A recent longitudinal study confirmed that daily cannabis use was related to better treatment outcomes [68]. Cannabis is a significantly less harmful substance than heroin [2, 69], is cheaper and because of a recent South African Constitutional Court ruling, the use of cannabis in a private space is now legal [70]. The reported levels of alcohol use was lower than expected, considering that alcohol is the most commonly reported substance of use among patients accessing substance use treatment services in the province [71]. An explanation for this could have been under-reporting of alcohol use.

This paper focuses on baseline and 6-month data. Data at the 12 month and 18 month intervals are also need to be collated and analysed, particularly in relation to the main findings, i.e. retention, drug use, and mental health. Qualitative research is needed to make sense of the facilitators and barriers to retention, as well as the participants’ health and drug use. It is also important that an analysis be conducted of the outcomes of low threshold opioid agonist treatment using buprenorphine and a cost effectiveness comparison in localised South African settings to inform policy and programming.